**2. Methods**

A literature search was performed in PubMed, Scopus, Web of Science, Cochrane Library databases and Embase. The search included papers published from 1 January 1993 to 1 March 2020. Only English studies were considered. All authors participated in the search process and in the critical analysis of selected publications. Keywords used were: "microbiota", "pancreas", "acute pancreatitis", "chronic pancreatitis", "pancreatic cystic neoplasms", "pancreatic ductal adenocarcinoma", "diabetes mellitus", "neuroendocrine tumors", "probiotic", "prebiotic", "synbiotic".

#### **3. Microbiota in Healthy Pancreas**

Recently, the pancreatic physiological functions have been shown to have an impact on intestinal microbiota and vice versa [34], by a cross-talking system, in health and disease; hence, it could be possible to talk about "microbiota-pancreas axis" [35].

In an important study, Sun et al. [36] demonstrated that pancreatic β-cells, in mice, produced cathelicidin-related antimicrobial peptide (CRAMP), a protein with antimicrobial activity through bacterial membrane permeabilization [37,38]. They noticed that pancreatic CRAMP expression was induced by SCFAs derived from the gu<sup>t</sup> microbiota, underling the cross-talking system. In a recent study, Ahuja et al. [39] used a mice model with pancreatic acinar cell–specific deletion of Orai1, a Ca2+ channel necessary for exocytosis of pancreatic antimicrobials. They found that Orai1-deficient mice showed an altered intestinal microbiota with bacterial overgrowth. In particular, *Proteobacteria* were two-fold increased, including increases in *Succinivibrionaceae* and *Enterobacteriaceae*, and *Prevotella* spp. The authors did not find differences on expression of intestinal antimicrobials in Orai1-deficient mice, proving the role of the pancreas in control of intestinal microbiota. Moreover, Orai1-deficient mice developed spontaneous intestinal inflammation characterized by intestinal CD3+ T-cell infiltration and died after three weeks. This data could be translated on humans, because patients that carry ORAI1 nonsense mutations have susceptibility to GI infection and diarrhea [40]. In an American cohort of patients who underwent a pancreaticoduodenectomy, Rogers et al., studied fecal microbiota composition compared to a group of healthy individuals. Interestingly, surgical patients exhibited fecal dysbiosis and showed an increase in *Klebsiella* and *Bacteroides*, while anaerobic taxa (i.e., *Faecalibacterium prausnitzii* and *Roseburia* species) decreased [41]. After DCP there are several anatomical alterations apart from the pancreatectomy (i.e., partial gastrectomy, the biliary-digestive anastomosis). In this scenario, also the gastric juice changes its composition. These data sugges<sup>t</sup> that both pancreatic and gastric juice, with their antimicrobial activity, are able to modify gu<sup>t</sup> microbiome.

#### **4. Microbiota in Pancreatic Diseases**

#### *4.1. Microbiota and Acute Pancreatitis*

Acute pancreatitis is caused by gallstones or acute alcoholic intake in the majority of cases. Rare causes of pancreatitis are hypertriglyceridemia, hypercalcemia, genetic causes, smoking, viruses such as cytomegalovirus or herpes simplex virus [42]. The main therapy for acute pancreatitis is early fluid resuscitation; the latter plays a key role by improving the tissue oxygenation and the microcirculation perfusion in order to preserve on the one hand pancreatic function (avoiding further pancreatic necrosis), and on the other hand, renal, cardiac and intestinal perfusion [43,44]. The use of antibiotics is not recommended routinely in severe acute pancreatitis (SAP), but only in the presence of infection, as cholangitis associated to biliary pancreatitis, and suspicion of pancreatic necrosis. In AISP (the Italian Association for the Study of the Pancreas) guidelines [45], the routine use of probiotics is not recommended. Besides, the use of probiotics in clinical practice is not cited in the recent American guidelines addressing the managemen<sup>t</sup> of pancreatic necrosis [42].

Acute pancreatitis determines hypovolemia that induces splanchnic vasoconstriction to preserve vital organs. Sometimes, this situation favors leaky gut. Indeed, high intestinal permeability promotes bacterial translocation and endotoxemia, with increased risk of infection of necrotic tissues [46]. Moreover, small intestinal bacterial overgrowth (SIBO) and bacterial translocation may be responsible for the majority of the pancreatic complications, but evidences are still scarce. Capurso et al. [44] showed that intestinal ischemia damages gu<sup>t</sup> barrier with bacterial translocation, endotoxemia and rarely multi-organ dysfunction syndrome.

Tan et al. analyzed differences in microbiome in patients with mild acute pancreatitis and SAP and a control group [47] (Table 2). They found fecal dysbiosis characterized by an increase in *Enterococcus* species and a decrease in *Bifidobacterium* species in SAP, when compared with other patients with mild form and control group. These data were in line with other studies [48].

In a recent case control study, Zhang et al. [49] found that fecal samples from 45 patients with acute pancreatitis contained fewer *Actinobacteria* and *Firmicutes* and more *Bacteroidetes* and *Proteobacteria* than those from 44 healthy controls.

The first randomized controlled trial (RCT), assessing the usefulness of probiotics on clinical outcomes of acute pancreatitis (AP), was conducted by Olah et al., in 2002 in Hungary [50] (Table 1).

The results were encouraging. Forty-five patients were recruited, 22 of them received *Lactobacillus plantarum* (*LP*) 299 with a substrate of oat fiber for 1 week by nasojejunal tube; instead, 23 patients were treated with a similar product with *Lactobacillus* inactivated by heat. Only one patient in the first group had pancreatic necrosis versus seven patients in the control group. Similar data was published by Qin et al. [48]. The authors enrolled 74 patients with AP and divided them into two groups: the first was treated with classical parenteral nutrition and the other was treated with enteral feeding and probiotics, containing living *LP*, administered through a nasojejunal tube. Analyzing fecal samples, they found that the second group was colonized with lower number of pathogenic organisms than the first one; besides, the first group had increased permeability with an increase of *Enteroccoccus* spp. and a reduction in *Lactobacteria* and *Bifidobacteria*. Thus, *LP* enteral feeding induced beneficial e ffects on maintaining the integrity of intestinal mucosal barrier and reducing the risk of septic complications. Olah et al., in 2007, published another RCT [51] with a new synbiotic composition, named "Synbiotic 2000", in the treatment of patients with AP. Sixty-two patients were randomized to probiotics plus prebiotics treatments or to prebiotics treatment alone. Patients in the first group had less complications (i.e., systemic inflammatory response, multiorgan failure) than the control group, but the di fference was not statistically significant. In the multi-center, nationwide, double blind placebo-controlled clinical trial called PROPATRIA study [52], 298 patients were enrolled. The probiotic group was treated with *Ecologic 641*, administered twice daily by nasojejunal feeding. It consisted of six strains of microorganisms (four types of *Lactobacilli* and two of *Bifidobacteria*). Instead, the control group received placebo. The authors found [53] that there was no di fference in the risk of infections between patients treated with probiotics and the control group (30% versus 28%). However, the data were dramatically di fferent concerning mortality. Indeed, 24 of 152 patients treated with probiotics (16%) died versus 9 of 144 (6%) in the placebo group. In particular, a high risk of mesenteric ischemia was seen in the probiotic group versus the control group. However, Bongaerts et al., made a reassessment of this study eight years later, analyzing the weak points of PROPATRIA trial [54]. One of them was the latency time in the first administration of probiotics; indeed, some patients were treated 24 h after onset of symptoms. Furthermore, there were errors in randomization; in fact, onset of multi-organ failure was already present during admission in more patients in the first group than in the placebo group (41 patients versus 23 patients).

Finally, they suggested, that future studies should evaluate the usefulness of high dose of probiotics, but caution is mandatory to prevent bacterial overgrowth.

Gou et al., in a meta-analysis published in 2014 [55], asserted that probiotics showed neither adverse nor beneficial e ffects on the clinical conditions of patients with SAP. However, they highlighted a significant heterogeneity in the studies of clinical outcomes. However, they highlighted a significant heterogeneity in the studies of clinical outcomes.

Currently, data are not su fficient to draw a conclusion regarding the e ffects of probiotics on patients with SAP, since they are controversial and heterogeneous. Future studies should improve the study designs, for example, detecting a peculiar strain of microorganisms (i.e., the type of probiotic), the correct dose of probiotics, or by a standardization of treatment duration.

#### *4.2. Microbiota and Chronic Pancreatitis*

Chronic pancreatitis (CP) is a persistent disease characterized by chronic inflammation of the gland with typical findings, such as Wirsung dilatation, intraparenchymal calcifications, gland atrophy, and development of fibrosis [56]. Chronic pancreatitis leads to progressive endocrine and exocrine dysfunction. Data sugges<sup>t</sup> that patients with CP display dysbiosis with growth of pathogenic

bacteria. This situation is related to the reduction of antimicrobial peptides in the pancreatic juice [57]. Capurso et al. analyzed by a systematic review and meta-analysis the relationship between SIBO and CP [58]. The authors proved that about 36% (range between 14–92%) of CP patients are affected by SIBO, detected by a glucose or lactulose hydrogen breath test. However, because of high heterogeneity in the results, the relationship is unclear, and further research is needed.

Pancreatic enzyme replacement therapy (PERT) is able to alleviate exocrine pancreatic insufficiency (PEI) related symptoms, helping fat absorption and improving nutritional status of patients [59]. On the other hand, some authors hypothesize that PERT could directly modify gu<sup>t</sup> microbiota, but the mechanism is still unknown. In the study by Nishiyama et al. [60], the authors evaluated the gu<sup>t</sup> microbiota from fecal samples in mice treated with PERT versus controls. Their findings supported the hypothesis that PERT alleviates PEI-associated symptoms by ameliorating digestive activity and by changing the composition of gu<sup>t</sup> microbiota. In particular they found a relative abundance of *Akkermansia muciniphila* and *Latobacillus reuteri* in mice treated with PERT.

In fact, researchers found a relative abundance of two types of microorganisms in fecal samples from pancrelipase-treated mice compared with controls, i.e., *Akkermansia muciniphila* and *Lactobacillus reuteri*. In particular, the first one, found 58-fold higher in the PERT group, is known to enhance intestinal barrier function by promoting mucus thickness and tight junction protein expression reducing the leaky gu<sup>t</sup> [61]; so, this study suggests that PERT may help to maintain *eubiosis*.

Recent studies highlighted the involvement of immune responses against intestinal microbiota in the development of CP [62]. For example, the activation of pattern recognition receptors (PRRs), i.e., toll-like receptors (TLR) and nucleotide-binding oligomerization-like (NOD) receptors that detect pathogen-associated molecular patterns (PAMPs) derived from the intestinal microbiota have been reported to play a critical role in the development of experimental CP [63].

Proton-pump inhibitors (PPI) are not recommended in patients with CP, but in refractory steatorrhea they could be helpful. In fact, patients with severe forms of inflammation can have lower bicarbonate secretion that cannot neutralize the acidity in the duodenum [64].

In a randomized, controlled, double-blinded trial, dos Santos et al., used synbiotics for changing intestinal microenvironment of patients with CP [65]. Synbiotics was composed of 12 g/day of *Lactobacillus casei*, *Lactobacillus rhamnosus*, *Lactobacillus acidophilus*, *Bifidobacterium bifidum* and FOS, and was administered to the intervention group; instead, 12 g/day of medium absorption complex carbohydrate was administered to the control group. The authors found that synbiotics improved clinical and laboratory outcomes in patients with CP. Dylag et al. [66] found that synbiotics may help restoration of the gu<sup>t</sup> microbiota through the fermentation of FOS and the consequent increase in *Bifidobacteria* and the reduction in pathogenic bacteria. The use of prebiotics and synbiotics in patients with CP promotes a reduction in intestinal stool frequency [65], but their clinical relevance role warrants additional investigation.

Collectively, these promising results sugges<sup>t</sup> that manipulation of microbiota, due to low cost and high manageability, may represent a new therapeutic frontier in CP.


**Table 1.** Randomized clinical trial about the use of probiotics, prebiotics or synbiotics in acute and chronic pancreatitis.

#### *4.3. Microbiota and Autoimmune Pancreatitis*

Autoimmune pancreatitis (AIP), described for the first time in 1995, is a rare fibro-inflammatory form of CP with typical imaging and histological findings, caused by autoimmune abnormality [67,68]. There are two subtypes of AIP: type 1 and type 2. Type 1 is the pancreatic manifestation of a systemic Immunoglobulin G4–related disease (IgG4-RD); it is associated with increased IgG4 serum concentrations, lymphoplasmacytic infiltrate and phlebitis in the pancreas. Instead, type 2 AIP affects only the pancreas, has normal IgG4 serum concentration and, histologically, shows a neutrophilic inflammation [69]. Hereditary and environmental factors together are thought to induce adaptive immune responses to self-antigens [70]. Microbial antigens may underlie the pathogenesis, but the trigger for the autoimmune cascade remains unknown.

At the beginning of the 2000s, several studies have examined the possible association between *Helicobacter pylori* (*H. pylori*) infection and pancreatic diseases, including AIP [71–74].

In 2010, Jesnowski et al., did not isolate *H. pylori* DNA either from samples or pancreatic juice of patients affected by AIP [75], excluding a direct bacterial infection of the gland. Afterwards, it has been proposed that *H. pylori* could cause AIP due to molecular mimicry mechanism. Guarneri et al. [76], found a significant homology between *H. pylori* a-carbonic anhydrase (a-CA) and human CA type II, enzyme highly expressed in pancreatic ductal cells. Moreover, the homologous CA segments contain the binding motif of the HLA molecule DRB1\*0405, which confers a risk for AIP development [76]. In 2009, Frulloni et al. [74] reported that 94% of AIP patients exhibited IgG antibodies to *H. pylori* plasminogen-binding protein (PBP), that is homologous to the human protein ubiquitin-protein ligase E3 component n-recognin 2 (UBR2), highly expressed in pancreatic acinar cells. Instead, this finding was found in only 5% of patients with pancreatic cancer adenocarcinoma (PDAC) and none of patients with alcohol-induced CP or intraductal papillary mucinous neoplasms (IPMN) [74]. Unfortunately, to date, these findings have not been confirmed. At first, Buijs et al. in 2016 [77], and then Detlfsen et al. [78] in 2018, did not find significant differences in serum concentrations of anti-PBP antibodies in AIP patients. Furthermore, in the second study the authors investigated the anti-CAII antibodies serum concentration, but no differences were found with the control group. Finally, in a prospective study, Culver et al. [79], found that the prevalence of gastric ulcerations, exposure to *H. pylori*, cytokine response and immunological memory to *H. pylori* PBP did not differ in IgG4-RD patients compared with control group. Hence, they ruled out a role for *H. pylori* PBP as a microbial antigen in IgG4-RD pathogenesis [79].

Watanabe et al., proved that antigens derived from intestinal microflora, through the activation of NOD-2 and TLR pathways, enhance IgG4 responses by peripheral blood mononuclear cells (PBMCs) from AIP patients, showing a possible involvement of innate immune responses against intestinal microflora in the development of AIP [80]. Several AIP-like experimental models have been induced in transgenic mice inoculating microbial agents: C57BL/6 mice infected with the murine leukemia retrovirus LP-BM5, developed histological findings similar to human AIP [81,82]; in MRL/Mp mice the administration of polyinosinic:polycytidylic acid (poly I:C), a synthetic double-stranded RNA and TLR3 ligand, promotes the development of AIP-like pancreatitis [83–86]. After showing in their first study that repeated inoculations with heat-killed non-pathogenic *Escherichia coli* (*E. coli*) (ATCC 25922), a common commensal bacterium from gu<sup>t</sup> microbiota, into C57BL/6 mice induced AIP-like pathological alterations accompanied by an elevation in serum IgG [87], Yanagisawa et al. identified [88] the *E. coli* antigen capable of these alterations: FliC, the major component of the flagella, an important cell surface structure of Gram-negative bacteria. In murine model, when injected intraperitoneally, FliC was able to induce AIP-like pancreatitis. Moreover, serum concentration of anti-FliC antibodies was found to be significantly higher in AIP patients than in those with CP, pancreatic cancer, and pancreatic disease-free controls. These findings indicate that FliC from *E. coli* may be involved in the pathogenesis of AIP [88]. Finally, Kamata et al. [89], investigated in a murine model of AIP the role of immune response against intestinal microflora, via plasmacytoid dendritic cells (pDCs) activation, whose products, interferon (IFN)-α and interleukin (IL)-33, are responsible

for chronic inflammation and pancreatic fibrosis [89–92]. They proved that bowel sterilization by broad spectrum antibiotics decreased pancreatic accumulation of pDCs, and consequently IFN-α and interleukin (IL)-33 levels, halting AIP development. They observed a reduction in the microbial diversity as well as alteration in the microbial composition in the feces of mice with AIP, with the abundance of *Bifidobacterium* spp. in the gu<sup>t</sup> of AIP mice. Lastly, they proved that intestinal dysbiosis, due to transmission of intestinal microflora by co-housing or by FMT, makes mice more sensitive to develop experimental AIP when they were injected with a lower dose of an inducing agen<sup>t</sup> [89]. If these data are confirmed in the human form of AIP, patients with this condition might be efficiently treated with a blockade of IFN-α and IL-33, in combination with the normalization of the intestinal microflora.

#### *4.4. Microbiota and Type 1 Diabetes Mellitus*

Type 1 diabetes (T1D) is an autoimmune disease characterized by the destruction of pancreatic insulin-producing β cells. Susceptibility to T1D is influenced by both genetic and environmental factors that act on immune system dysregulation [93]. The risk of developing T1D, in fact, has been related to viral infections, dietetic factors, and vitamin D deficiency, while the role of antibiotics is still controversial [94].

The impact of intestinal microbiota in T1D predisposition and pathogenesis has been studied in human subjects and in murine models, such as non-obese diabetic mice [95,96]. Early-life events, such as delivery mode, breastfeeding, solid food introduction and antibiotics, strongly influenced the microbiota composition and its interaction with mucosal and systemic immunity [95]. Diabetes resulted to be associated with higher *Bacteroidetes*/*Firmicutes* ratio and lower α-diversity in fecal samples [95,96]. Moreover, subjects affected by T1D showed an increase in levels of *Clostridium*, *Bacteroides dorei* and *vulgatus*, *Blautia*, *Rikenellaceae*, *Ruminococcus* and *Streptococcus*, while *Lactobacillus*, *Bifidobacterium* and SCFAs-producing bacteria were reduced [94,95].

Microbiota dysregulation affects the risk of T1D through different mechanisms, including intestinal permeability [97,98], molecular mimicry [99] and immune system modulation [100–103].

The intestinal barrier prevents the translocation of toxins, food antigens and infectious factors, which can activate the immune system. Specific microorganisms of gu<sup>t</sup> microbiota, including *Dialister invisus*, *Bifidobacterium longum*, *Gemella sanguinis* and *Clostridium perfringens*, can compromise the intestinal barrier, leading to an increased risk of T1D [94]. Alterations of the barrier function, in fact, are associated with activation and proliferation of islet-autoantigen specific CD8+ T cells, with pro-diabetogenic effect [98].

Moreover, children affected by T1D showed a significantly higher intestinal permeability [98].

The molecular mimicry is an important pathogenetic mechanism not only for diabetes but also for other autoimmune diseases. The molecular structure of some bacterial proteins showed to be similar to pancreatic self-antigens. Monofunctional glycosyltransferase protein (MGT) of *Leptotrichia goodfellowii*, for example, showed similarity to islet specific glucose-6-phosphatase related protein and could cause immune cross-reactions [94,104,105].

Innate immunity involves macrophages, granulocytes and natural killer cells, and a key role in the antigen recognition is played by TLRs, which are activated by PAMPs. Many components of intestinal bacteria, such as lipopolysaccharides (LPS), lipoproteins, peptidoglycan and nucleic acids, act like PAMPs and can activate TLRs, with pro-diabetogenic or anti-diabetogenic effects [101]. Different TLRs, in fact, can promote or inhibit autoimmune processes: TLR2-pathway showed to facilitate T1D, while TLR4-signaling resulted to be protective against diabetes development [101]. In addition, some microorganisms can modulate T-cells of B-cells functions. Even if the exact mechanism is still unknown, *Listeria* can induce Th1 response [102], while *Clostridia* can induce regulatory T cells [102]. SCFAs are associated with reduced serum levels of pro-inflammatory cytokines which promote T1D development, such as interleukin-21 (IL-21). They preserve the intestinal barrier function and promote B cells differentiation into plasma-cells and memory cells [103]. Even if there is no current evidence-based treatment for preventing or delaying T1D onset by acting on gu<sup>t</sup> microbiome, the above-mentioned

data sugges<sup>t</sup> some possible therapeutic approaches. The immune dysregulation, for instance, can be modulated through the administration of *Escherichia coli Nissle* (*EcN*), which is associated with reduced pathogenic bacteria colonization, decreased serum levels of IL-2 and TNF α and increased IL-10 [95]. Other possible strategies include the correction of dysbiosis and the improvement of microbial diversity through probiotics and prebiotics, such as inulin and oligosaccharides [95]. These treatments have been tested in mouse models, but further studies are needed to prove their e fficacy in humans.

#### *4.5. Microbiota and Pancreatic Cystic Neoplasms*

Pancreatic cystic neoplasms (PCN) represent up to 5% of the total amount of pancreatic cancerous neoplasms [106]. Frequently, the discovery of pancreatic cysts is casual, and these are found during screening or diagnostic procedures for other problems. Diagnostic refinement by computed tomography (CT) and magnetic resonance imaging (MRI) have allowed us to define PCN, undetected in the past. PCN are divided into various types, such as serous cystadenomas (SCA), mucinous cystic neoplasm (MCN) and IPMN; other forms of PCN are very unusual [107]. IPMN can be divided into three types: IPMN main duct, IPMN branch duct and the mixed form [108]. In recent years, researchers tried to find a connection between gu<sup>t</sup> microbiota and cystic pancreatic lesions, but some questions have not been answered yet. Following data reporting the correlation between poor oral health and PDAC [109,110]. Olson et al., in a pilot study, assessed the di fference between oral microbiota in patients with PDAC, IPMN and controls [111]. The authors evaluated the characteristics of the oral microbiota of 40 PDAC patients, 39 of whom with IPMN and 58 patients included as control group. Eligible participants were age 21 or over, had not smoked tobacco products in the past year, had not taken antibiotics in the past 30 days, had not been treated for any cancer (other than non-melanoma skin cancer) in the past two years. The 16s rRNA microbiota screening was performed in saliva samples. Patients with PDAC had a higher mean relative proportion of *Firmicutes*/*Bacteroidetes* ratio, while *Proteobacteria* were predominant in controls; in fact, the first group had high levels of *Firmicutes* and *Lactobacillae*. However, di fferences were observed only in the main relative proportion of some Operational taxonomic unit (OTU). No di fferences were detected between PDAC and IPMN groups about alpha diversity [111]. Gaiser et al., have reported a prospective study about the harbor of intracystic microbiome in PCN [112]. Patients, undergoing pancreatectomy, were enrolled. Cystic fluid samples were collected at the day of surgery. The authors found that 16s DNA copy number were higher in IPMN with high grade of risk malignancy than benign neoplasms. Moreover, IL-1 beta protein was higher in IPMN and cancer. In particular, *Proteobacteria* were predominant in benign cystic forms, while *Firmicutes* in IPMN with high grade dysplasia (HGD) and PDAC. A sub-analysis was performed to investigate the potential role of invasive endoscopic procedures [i.e., endoscopic ultrasonography-fine needle agoaspiration (EUS-FNA) or endoscopic retrograde cholangiopancreatography (ERCP) or percutaneous transhepatic cholangiography (PTC)] in the microbiota composition. *Fusobacterium nucleatum* (*F. nucleatum*) was one of the predominant pathogens in cystic specimens. qPCR assay confirmed that *F. nucleatum* were increased in IPMN HGD and in PDAC. Previous procedures had no impact on microbiota composition, as well as the previous use of PPI [112]. It is now believed that *F. nucleatum* have an oncogenic role in some cancers, such as colorectal cancer [113] and PDAC itself [114].

Li et al. [115] studied pancreatic cys<sup>t</sup> fluid (PCF) to determine the genera of bacteria present. Sixty-nine pancreatic fluid samples were collected. Twenty-seven were IPMN, 13 MCN, nine pseudocysts and nine SCA; the rest were classified as "other". Predominant microorganisms in the cysts were *Bacteroides* spp. and *Escherichia* spp., and even less *Fusobacterium* spp. and *Bacillus* spp. The authors found that PCF contained their bacterial microenvironment with a unique ecosystem. Finally, *Helicobacter* were marginally detected in pancreatic cys<sup>t</sup> fluid. In this case, more studies are necessary as may be further warranted.

#### *4.6. Microbiota and Pancreatic Ductal Adenocarcinoma*

Pancreatic ductal adenocarcinoma is the 12th most common cancer worldwide [116] and the fourth most fatal cancer in both men and women [117]. The incidence of PDAC is higher in North America and in Western Europe and it is more common in men than in women [116].

Risk factors for PDAC include cigarette smoking, obesity and CP. Some chemical compounds, nickel-based, chromium-based, and silica dust have been reported to increase the risk of PDAC too [117]. Moreover, the development of PDAC has been related to dietary factors, *H. pylori* infection, oral, gu<sup>t</sup> and pancreatic microbiota [117]. The mechanisms by which microbiota influences carcinogenesis involve innate and adaptive immune suppression [118,119] and stimulation of pro-carcinogenic cellular pathways [118,120]. In preclinical mice models, in fact, microbial ablation with broad-spectrum antibiotics has been associated with increased presence of intratumoral T cells [117], while the administration of cell-free extracts from fecal samples of gu<sup>t</sup> bacteria from PDAC-bearing mice or cell-free extracts from *Bifidobacterium pseudolongum* were associated with lower expression of MHC II and up-regulation of IL-10 [118]. Moreover, bacteria-associated PAMPs bind to specific TLRs, such as TLR4, thus activating MAP (mitogen-activated protein) Kinase and NF-kB (nuclear-factor kappa-light-chain-enhancer of activated B cells) pathways, which are potent promoter of carcinogenesis [121].

The etiologic relationship between gu<sup>t</sup> microbiota and PDAC has been initially studied in mice models. In these studies, germ-free mice and those treated with oral antibiotics showed a lower incidence of pancreatic cancer, while fecal transplant from PDAC-bearing mice resulted in an increased risk of tumor [118].

In murine models, the presence of PDAC has been related to increased number of gu<sup>t</sup> bacteria from fecal samples belonging to *Bacteroides*, *Firmicutes* and selected genera associated to *Actinobacteria* and *Deferribacteres*. Moreover, different microbiota composition has been related to a different rate of pancreatic carcinoma progression [118]. *Elizabethkingia*, *Enterobacteriaceae* and *Mycoplasmacetaceae* resulted to be associated with slower progression, while aggressive PDAC was associated with *Helicobacteriaceae*, *Bacteroidales* and *Mogibacteriaceae* [118]. In humans, *Proteobacteria*, *Actinobacteria*, *Fusobacteria* and *Verrucomicrobia*, which normally represent a minor proportion of human intestinal microbiome [122], were present in abundance in the gu<sup>t</sup> of PDAC patients if compared to controls [118].


**Table 2.** Microbial changes in the pancreatic diseases.

intestine bacterial overgrowth; T1D: type 1 diabetes.

In epidemiological studies, poor oral health and periodontal diseases have been related to PDAC incidence [121]: these data sugges<sup>t</sup> a possible relationship between oral microbiome and pancreatic cancer, as mentioned above. In human studies, in fact, the carriage of *Porphyromonas gengivalis* and *Aggregatibacter actinomycetemcomitans* and a decreased proportion of *Fusobacteria* and its genus *Leptotrichia* resulted to be associated with higher risk of PDAC development [110]. Many studies have demonstrated the role of *Porphyromonas gingivalis*, an oral pathogen that causes periodontitis and gingivitis, in the PDAC. Smoking and alcohol consumption are recognized risk factors for pancreatic cancer, but at the same time they can a ffect oral microbiome composition. Therefore, the association between oral bacteria and PDAC has been studied in ever and never-smokers and in ever and never-drinkers: the results sugges<sup>t</sup> that the association between oral microbiota and PDAC are not likely due to these potential confounders [110]. The presence and composition of pancreatic microbiota have been associated to PDAC development: cancerous pancreas showed a more abundant microbiota compared to normal pancreas in both murine models and human subjects [118]. Translocation of gu<sup>t</sup> bacteria and their migration into the pancreas have been demonstrated by Pushalkar et al. [118], through comparison between fecal samples and pancreatic microbiota and through oral administration of fluorescently labelled *Enterococcus faecalis* and Green Fluorescent Protein (GFP)-labeled *Escherichia coli*. Thus, gu<sup>t</sup> microbiota could influence pancreatic microenvironment. The composition of pancreatic microbiota in PDAC resulted to be di fferent from normal human pancreas. Thirteen phyla were detected in tumor tissue. The most abundant were *Proteobacteria*, *Bacteroidetes* and *Firmicutes* [118]. Interestingly, *Proteobacteria* only amount for about 8% of gu<sup>t</sup> bacteria from fecal samples in PDAC patients, while they were one of the most represented bacteria in their pancreatic tissue [118]; this suggests a di fferentially increased translocation of these Gram-negative bacteria to the pancreas. Nowadays, it is believed that *F. nucleatum*, an inhabitant of the oral mucosa, is important in the colon rectal dysplasia and cancer. Similar results were found on PDAC. In a landmark study, Mitsuhashi et al. [114] studied the correlation between PDAC and oral microbiota; they analyzed a database of 283 patients with PDAC who underwent surgery and found *Fusobacterium* spp. in 8.8% of pancreatic cancer samples; the presence of tumor *Fusobacterium* was not associated with any molecular and clinical features, but with worse prognosis. Mitsuhashi et al. suggested that *Fusobacterium* could be a negative prognostic negative independent biomarker for PDAC.

This data about intestinal, oral and pancreatic microbiota can represent useful screening and prevention tools. In fact, they sugges<sup>t</sup> that the analysis of oral and gu<sup>t</sup> microbiota composition could be used for risk stratification and early diagnosis of PDAC. Moreover, oral antibiotics could be proposed as a preventive therapy for high-risk patients or in association with chemotherapy with a synergic effect [118]. Both these diagnostic and therapeutic implications, however, need further studies to be confirmed.

#### *4.7. Microbiota and Pancreatic Neuroendocrine Tumors*

Evidences about a link between pancreatic neuroendocrine tumors (pNET) and microbiota are very poor. Currently, therapies for pNET are surgery for not invasive neoplasms, somatostatin analogues (SSA), peptide receptor radionuclide therapy (PRRT), target therapy with Everolimus and Sunitinib and chemotherapy (for advanced NET) [123]. In this scenario, immunotherapy could have a futuristic role; in particular, PD-1 inhibitors and PD-1 ligands are a group of checkpoint inhibitors that have been developed for treatment of some cancers (i.e., melanoma, squamous cell lung cancer, advanced cell renal carcinoma, Merkel-cell carcinoma, a cutaneous neuroendocrine tumor) [124,125]. Specific clinical trials are looking for a role of immune checkpoint inhibitors in pNETs [126], that may conduct to more personalized target therapy; in particular researchers are testing pembrolizumab in NET with advanced stage. In a landmark study of Nghiem et al. [125], they used pembrolizumab as first-line therapy in patients with advanced Merkel-cell carcinoma (a type of skin neuroendocrine tumor) with an objective response rate of 56%.

Recent data support a role for the commensal microbiota in the efficacy of immunotherapy itself. Indeed, gu<sup>t</sup> microbiota may have a "mechanistic impact" on antitumor immunity in the cancer [127]. Matson et al., collected 38 stool samples from metastatic melanoma patients on anti-PD-1 treatment and, after 16S RNA sequencing and quantitative PCR analysis, they identified *Bifidobacterium* spp., *Colinsella aerofaciens*, *Enterococcus faecium* as bacteria associated with beneficial response [127]. They transplanted fecal material from responding patients in a germ-free mice model of melanoma. They observed that FMT could lead to improve tumor control, augmenting T cell responses, and greater efficacy of anti-PD-L1 therapy.

The commensal microbiota composition might be useful as a biomarker to predict response to checkpoint blockade therapy, and in the future, it could be possible to translate this research for the pNET. However, data are very poor now, and the relationship between immunotherapy and microbiota, in NET patients, is still unclear.
